Forum Discussion

johnbor's avatar
johnbor
Connection Cadet
3 years ago

Speed Drops the regains and restart ?

I have had the T-Mobile trash can for about a year and my speeds have always been kind of slow like 30 -60 down and 7-9 up, now that’s fine for me but noticed this past week the speeds have changed . they are fine for about 1-2 hours into using the internet during OFF PEAK or ON PEAK times.  Then the speeds would drop to 3-8 down and 7-8 up. thanks right the up speed stays the same and this will continue BUT the minute I RESTART the trash can and wait the 2 minutes for it to come up the speed returns to normal for another 1-2 hours then it all begins again. Now like I said this just started this week and never did this in the past year.

Any ideas ?

 

  • Curious that the behavior only recently started. When the behavior is such that the download have dropped down it would be good to know a couple of things. You have the Nokia gateway as you lovingly refer to it as a trash can. :-) Well, it sort of resembles one but compared to the other two newer ones I am not sure it is so bad myself. Mine has been a good little can so I am not convinced they are all bad. 

    Let me get to the point. The Nokia is so simple to get the cellular metrics from and also get the PCI values for the primary and secondary cellular signals that is where we should start. I suggest to take some time and focus on the cellular metrics and record what your gateway reports. On the overview page it reports the cellular signal metrics for both the primary and secondary frequencies. When the speed is up or down record these and compare good to bad. Is there a change is what I am after. Also the same needs to be done with the cellular information on the Status page for the cellular network information for both primary and secondary information. Paying attention to the PCI value for both but mainly for the 5G NR secondary signal source. Does the PCI value change or is it the same. Does the PCI value change after the reboot of the gateway for one or both or either signals after the reboot? Does your gateway receive the n71 or the n41 5G cellular frequency or flip back and forth?

    The reason I ask about the PCI, physical cell identifier is that it could be when the speed is so reduced the secondary 5G cellular source is NOT the same as when it was 10X before the degredation. The PCI relates to the cell that the signal source originates from. If you use CellMapper.net to locate the cells that are related to the reported PCI values you can determine exactly where the tower is that has that cell. The 4G LTE and the 5G NR cells might be on the same tower or may not be but if the gateway does see another 5G cellular signal that is stronger than the one it had and it meets the threshold to move then it may lock onto the “new” signal source and have different behavior. The move may or may not make sense but it might happen. It sort of depends upon your location with respect to different cellular signals. 

    The speeds you get for download are not very impressive for 5G but well more like say a DSL connection which I have had and it is rather meh… so not trying to be too critical but I would expect better for 5G. With that said the speeds possible can and will depend on multiple variables and if you are out on the edge of the reach/range of the 5G signal source that would explain why it is slower. 

    The performance is NOT just about speed. If you have a strong signal that is a quality signal and has low signal to noise ratio then it will perform better. A signal with good strength and poor quality and noise is still probably going to perform bad as noise and a poor quality signal will lead to packet loss and retransmissions due to packet loss or damage. So, what i am saying is the metrics matter and you have the gateway that has those tools easily available to reach and use.

    You can combine the observations of the cellular metrics with speed testing with say speedtest.net and fast.com to profile the behavior. I prefer to run speed testing with my MacBook Pro on a wired Ethernet cable vs using the wireless. My MacBook Pro is from 2015 so not 802.11ax capable. Older network adapters will not be able to perform as well as the new ones and can influence the testing. With speedtest.net I suggest to download the application, create an account, and then configure the preferred server as a baseline to test against. Test against a number of test servers to begin with and find one or two that seem to provide positive results. Pick one you like and use it as a reference. The more consistent your testing is the better the profile. Once you have the account you have a historical record of the testing so you can have a clear record of how the behavior is or was. If you make screen clips of the information and keep up with things you can build a good profile and really know what to expect. 

    I know it seems a bit excessive but it helps. Knowing where the tower is that has the cellular source really helps to get the gateway in a better placement. Don’t just settle for a couple of bars and it works. Try rotating the gateway clockwise or counter clockwise and see if the signal wash over one side of the can or the other raises the metric values and improves the operation. I have my gateway in a window in the basement with exposure somewhat to a tower 5.3 miles north of our home and I spin it around so that the back side, where the Ethernet cables plug in are facing in. Sure it is ugly but it improves the signal wash over the 5G antennas on the “front” side of the gateway by -3 to -5 dBm for RSRP and usually improves RSRQ and the SINR by a db or two. You may understand the metrics so please forgive me for being verbose on the matter. I will include the chart I use for reference.

    Hopefully the information will help you tune it up so you can get a better, stronger signal. Your downloads are much lower than I would hope for given it is 5G but once you have a clear picture of what the cellular metrics are telling you I am sure might be able to gain some improvement. I say this but IF he distance to the tower puts you on the edge of the cell reach for 5G the only recourse for that is to use/add an external antenna. The external antennas are not cheap but they can do wonders for improving the signals. If both the 4G and 5G signals originate on the same tower establishing a 2X2 or 4X4 MIMO antenna, depending on the 5G frequency, can improve things. 

    You asked for ideas. OK, so there you go. Probably more than you wanted to know but it is cellular and not a cable solution so it takes more effort and work. 

  • johnbor's avatar
    johnbor
    Connection Cadet

    My values and metrics dont change thats the funny part they stay the same whne its fast and then when its slow but one thing IKNOW is the restart of the modem for the 2 minutes fixes the problem for another hour or two.

    Is there a way to see a possible log?  to find why all the sudden this is occuring?

  • T-Mobile does not have any baked in log of the gateway functions as far as I know. Since they have destroyed the use of ICMP using the ping utility and/or trace route is not an option. 

    One thing you can do, that might be helpful, is to check your external IPv4 and IPv6 address information by going to the site, https://whatismyipaddress.com and checking before and after the change in behavior. Does it still show your location in the same city? I checked mine and it shows up in Nashville though I am in eastern TN 200+ miles away. If the location changes that might be a bit of a clue. Maybe it does not change where the traffic surfaces. Just looking for things to check and see what may change. 

    It could be something related to traffic handling in the local tower equipment. If that is the case the reboot might be the only workaround for now. I don’t know quite what it might be other than a buffering issue. 

    One thing you could do if you are somewhat technical is run a Wireshark packet capture and examine the traffic flows. You can simplify it by using the Analyze option > Expert Information. This will pop a window with a listing of packets and you can see what is going on. If you run a capture before when things are going good and then after the change it might present a bit different profile in the traffic handling. If there is a significant increase in retransmissions and “warnings” that might be helpful to know. 

    I ran a packet capture for maybe 15 seconds and could see a couple of things I did not expect but one DNS thing was related to my iPhone so that seems to be a bit one off. I don’t see an indication of traffic problems currently and my download is excellent today even at noon. Ever since they changed our signal source from n71 to n41 I have been keeping an eye on things. It was super stable on the n71 so when the transition to n41 took place it made me a bit anxious as I have seen those conversations in the community about how the n41 cells are stood up and at first they are great and then choked down to pretty much near useless. I suspect there is some overloading in some urban markets where expansion is planned but not implemented broad enough so too many users are on the cells leading to congestion and excessive throttling. Given your speeds either you are on the edge of the cell or the signals are just not as good as one would like. Without knowing the cellular metrics it is impossible to know.

  • The main culprit I suspect is buffer bloat and congestion. When looking at packets and the TCP information looking for the explicit congestion notification information is a dead give away. If there is no active queue management then the ECN will not take place. 

    Buffer bloat, overflow, packet loss, poor performance = issue

    You can go to Wikipedia and just search for, “explicit congestion notification” and the information is there.

    OR > https://en.wikipedia.org/wiki/Explicit_Congestion_Notification

    Of course understanding IP packet structure is part of the game. 

    ECN uses the two least significant (right-most) bits of the Traffic Class field in the IPv4 or IPv6 header to encode four different code points:

    • 00 – Non ECN-Capable Transport, Non-ECT
    • 10 – ECN Capable Transport, ECT(0)
    • 01 – ECN Capable Transport, ECT(1)
    • 11 – Congestion Encountered, CE.

    In my packet captures so far I have only seen some indications of damaged or dropped packets leading to retransmissions. 

  • Regarding the https://whatismyipaddress.com/ page. You can go there in a browser tab and it will report your IPv4 and IPv6 address pretty much after the page renders. If you leave the page up and you see the behavior change where things slow down just go to the page and reload the page. If the IP address reported has changed well, maybe that is a clue. Then do the router reboot and after the reboot go back to the page and see what the IP address is after the reboot. It will most likely be the same, I would think, but it is something you can check. The problem might be congestion due to buffer bloat and the reboot sort of resets things clearing out buffers. An IP address change may have no bearing on the behavior.

    If you were to check the IP address and see a pattern that emerges then you might have something.

  • johnbor's avatar
    johnbor
    Connection Cadet

    I reboot the T-Mobile Trash can not my computer… Solutions ?

  • johnbor's avatar
    johnbor
    Connection Cadet

    Speeds now are the fastest ever, something going on My secondary signal strength increased and now Im getting the best speeds ever they had to do something on the tower.

    Is there a way to look at the metrics to find what tower Im hooked (connected) up to?

  • If it is a new n41 it may not be in the CellMapper.net database yet. It depends on the gateway you have. The Nokia reports the band and PCI of the cell. You might be able to get the tower location from support. 

  • OK so maybe we have something. I wonder about that tower. The speeds you had were more indicative of 4G LTE only NOT 4G & 5G both. I have to question if you had both the primary and secondary frequencies. If they reported they were working on the tower they may have been working on what was one of the prior Sprint towers and adding the n71 cell equipment. If the speeds have jumped up in a significant way that would explain the difference. The B2 primary is 4G LTE and the n71 secondary is the 5G NR gigahertz frequency. It is capable of pretty good speeds. Not like the n41 but in good conditions and being not too far away you might get 150-200 Mbps downloads. The B2 will primarily be handling uploads. I am just speculating but it is probably because the secondary was never right and they fixed it or they added the n71. Run speedtest.net and check your speeds and ping latency. It sounds promising.